Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 81 of the 81 references it cites were found in a registry or an index; none of the 50 sources its claims cite is checked for support yet.

From enhancement to over-reliance: a mixed-method study of generative AI and sustainable learning performance

Open access
Gao, Li, Sun, Yanli, Khan, Safeer Ullah · 2026 · Frontiers in Psychology

Abstract

The rapid integration of generative artificial intelligence (AI) in education has raised important questions regarding its impact on student learning outcomes. While prior research has primarily focused on AI adoption, limited attention has been given to its influence on sustainable learning performance and the potential risks associated with its use. Addressing this gap, this study develops a comprehensive model to examine both the positive and negative effects of AI use on learning by integrating AI literacy, self-regulated learning, cognitive offloading, and individual differences. A mixed-method approach was employed, combining time-lagged survey data collected in three waves from 623 university students in China with qualitative interviews conducted with educators. The quantitative data were analyzed using PLS-SEM, while qualitative data were examined through thematic analysis to provide deeper insights into the observed relationships. The findings reveal that AI literacy significantly enhances critical AI evaluation, which, along with self-regulated learning, promotes effective AI use. Effective AI use positively influences sustainable learning performance but also increases AI over-reliance, which negatively affects learning outcomes. Furthermore, polychronicity moderates key relationships, indicating that multitasking tendencies shape both AI dependency and learning effectiveness. The qualitative findings support and explain these results, highlighting the behavioral mechanisms underlying AI-supported learning. Overall, the study demonstrates that the impact of AI on learning is dual in nature and depends on how students engage with the technology, offering important implications for theory and educational practice.

Overview

CoChat Summary

TL;DR

This mixed-method study models both benefits and risks of generative AI for learning, finding that AI literacy and self-regulated learning support effective AI use, which improves sustainable learning performance but also increases AI over-reliance. Polychronicity moderates both pathways, strengthening the link from effective AI use to over-reliance and weakening its link to sustainable learning performance.

Key findings

  • AI literacy strongly predicted critical AI evaluation (β = 0.893, t = 111.167, p < 0.001, f² = 3.916), and critical AI evaluation plus self-regulated learning predicted effective AI use (β = 0.567 and β = 0.399, respectively, both p < 0.001).
  • Effective AI use positively predicted sustainable learning performance (β = 0.871, t = 30.115, p < 0.001, f² = 1.139) but also positively predicted AI over-reliance (β = 0.612, t = 25.574, p < 0.001, f² = 1.160); AI over-reliance negatively predicted sustainable learning performance (β = −0.205, t = 5.291, p < 0.001, f² = 0.044).
  • Polychronicity moderated effective AI use → AI over-reliance (β = 0.261, t = 10.422, p < 0.001, f² = 0.227) and effective AI use → sustainable learning performance (β = 0.113, t = 4.452, p < 0.001, f² = 0.036).
  • Model explanatory power: R² = 0.797 for critical AI evaluation, 0.865 for effective AI use, 0.679 for AI over-reliance, and 0.694 for sustainable learning performance; Q² values ranged from 0.360 to 0.626.

Methods

  • Explanatory sequential mixed-method design: quantitative phase first, followed by qualitative educator interviews to explain and contextualize findings.
  • Quantitative sample: three-wave time-lagged survey of Chinese university students; 1,000 questionnaires distributed in Wave 1, 771 valid responses, 693 in Wave 2, and 623 matched responses in Wave 3 (overall retention approximately 62.3%).
  • Analysis: PLS-SEM using SmartPLS 4 with 5,000 bootstraps for direct and moderation paths; qualitative data from 15 educators were analyzed using Braun and Clarke’s six-phase thematic analysis, then integrated via triangulation, joint display analysis, and meta-inference.

Limitations

  • The paper states that participants were drawn only from Chinese universities, limiting cross-cultural and cross-system generalizability.
  • The paper states that the design was ultimately cross-sectional and non-experimental despite the time-lagged waves, so causality cannot be firmly established.
  • The paper states that sustainable learning performance was measured only by student self-report, with no objective measures such as GPA, standardized retention assessments, or observable behavior; it also notes possible common method bias despite Harman’s test and full collinearity VIF diagnostics.
  • The text makes evident additional measurement concerns: the HTMT for effective AI use and sustainable learning performance was 0.859, close to the conservative 0.85 threshold, and sustainable learning performance had a relatively low AVE of 0.534.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3389/fpsyg.2026.1847369

Cites

81 of 81 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Conceptual model.
Figure 1 Conceptual model.CC BY · PubMed Central
Figure 2 Structural model.
Figure 2 Structural model.CC BY · PubMed Central
Figure 3 Moderation of polychronicity.
Figure 3 Moderation of polychronicity.CC BY · PubMed Central

Verification

Claims the paper makes

0 Valid · 8 Not valid · 47 Not checked · 156 Own results and interpretations

67 more claims

Create a free account to see all 211 claims and to verify the 48 references still open.

See all claims — free account
  • The quote only partly states this claim. 5
  • It may state two separate claims as one. 74
  • Probably not a substantial claim. 88

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

5 of 6 · the rest aren’t loaded

1 more finding

Create a free account to see every finding of this check.

See all findings — free account

Citations

Cites

Show all 80 references

Research works printed but not linked: 1

  1. Kline, 2023, Principles and Practice of Structural Equation Modeling, 5th edn

Metadata

Authors
Gao, L., Sun, Y., & Khan, S. U.
Affiliations
  • Student Aid Management Center, Jiangxi Institute of Technology
  • Weifang University (CN)
  • Shenzhen University (CN)
Corresponding author
Sun, Yanli
Venue
Frontiers in Psychology
Published
2026-06-29
Volume
17
Pages
1847369-1847369
ISSN
1664-1078 (ISSN-L)
OpenAlex
W7166520030
Cited by
1
References
80
Citations by year
2026: 1
FWCI
3.61
Type
journal article
Language
en
Open access
Locations
  • Frontiers in Psychology (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
Topics
  • Social Sciences › Education › Artificial Intelligence in Education
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Physical Sciences › Computer Science › Artificial Intelligence › Intelligent Tutoring Systems and Adaptive Learning
Keywords
generative AI in education, overreliance on AI, AI literacy, student self-regulation, cognitive offloading, individual differences, polychronicity, multitasking, university students, AI-assisted learning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-18, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-18 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: From enhancement to over-reliance: a mixed-method study of generative AI and sustainable learning performance. https://app.cochat.ai/papers/10.3389/fpsyg.2026.1847369
The paper
Gao, L., Sun, Y., & Khan, S. U. (2026). From enhancement to over-reliance: a mixed-method study of generative AI and sustainable learning performance. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1847369

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.